| Age | Commit message (Collapse) | Author |
|---|
|
[ Upstream commit 104a0c02e8b1936c049e18a6d4e4ab040fb61213 ]
On ThunderX T88 pass 1.x through 2.1 parts, broadcast TLBI
instructions may cause the icache to become corrupted if it contains
data for a non-current ASID.
This patch implements the workaround (which invalidates the local
icache when switching the mm) by using code patching.
Signed-off-by: Andrew Pinski <apinski@cavium.com>
Signed-off-by: David Daney <david.daney@cavium.com>
Reviewed-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Sasha Levin <alexander.levin@verizon.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
commit 04a848106193b134741672f7e4e444b50c70b631 upstream.
As reported by Zijun, the fdt_check_header() call in __fixmap_remap_fdt()
is not safe since it is not guaranteed that the FDT header is mapped
completely. Due to the minimum alignment of 8 bytes, the only fields we
can assume to be mapped are 'magic' and 'totalsize'.
Since the OF layer is in charge of validating the FDT image, and we are
only interested in making reasonably sure that the size field contains
a meaningful value, replace the fdt_check_header() call with an explicit
comparison of the magic field's value against the expected value.
Reported-by: Zijun Hu <zijun_hu@htc.com>
Acked-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
commit 2ce39ad15182604beb6c8fa8bed5e46b59fd1082 upstream.
Clearing PSTATE.D is one of the requirements for generating a debug
exception. The arm64 booting protocol requires that PSTATE.D is set,
since many of the debug registers (for example, the hw_breakpoint
registers) are UNKNOWN out of reset and could potentially generate
spurious, fatal debug exceptions in early boot code if PSTATE.D was
clear. Once the debug registers have been safely initialised, PSTATE.D
is cleared, however this is currently broken for two reasons:
(1) The boot CPU clears PSTATE.D in a postcore_initcall and secondary
CPUs clear PSTATE.D in secondary_start_kernel. Since the initcall
runs after SMP (and the scheduler) have been initialised, there is
no guarantee that it is actually running on the boot CPU. In this
case, the boot CPU is left with PSTATE.D set and is not capable of
generating debug exceptions.
(2) In a preemptible kernel, we may explicitly schedule on the IRQ
return path to EL1. If an IRQ occurs with PSTATE.D set in the idle
thread, then we may schedule the kthread_init thread, run the
postcore_initcall to clear PSTATE.D and then context switch back
to the idle thread before returning from the IRQ. The exception
return path will then restore PSTATE.D from the stack, and set it
again.
This patch fixes the problem by moving the clearing of PSTATE.D earlier
to proc.S. This has the desirable effect of clearing it in one place for
all CPUs, long before we have to worry about the scheduler or any
exception handling. We ensure that the previous reset of MDSCR_EL1 has
completed before unmasking the exception, so that any spurious
exceptions resulting from UNKNOWN debug registers are not generated.
Without this patch applied, the kprobes selftests have been seen to fail
under KVM, where we end up attempting to step the OOL instruction buffer
with PSTATE.D set and therefore fail to complete the step.
Acked-by: Mark Rutland <mark.rutland@arm.com>
Reported-by: Catalin Marinas <catalin.marinas@arm.com>
Tested-by: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Reviewed-by: Catalin Marinas <catalin.marinas@arm.com>
Tested-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
commit 0106d456c4cb1770253fefc0ab23c9ca760b43f7 upstream.
Commit 66dbd6e61a52 ("arm64: Implement ptep_set_access_flags() for
hardware AF/DBM") ensured that pte flags are updated atomically in the
face of potential concurrent, hardware-assisted updates. However, Alex
reports that:
| This patch breaks swapping for me.
| In the broken case, you'll see either systemd cpu time spike (because
| it's stuck in a page fault loop) or the system hang (because the
| application owning the screen is stuck in a page fault loop).
It turns out that this is because the 'dirty' argument to
ptep_set_access_flags is always 0 for read faults, and so we can't use
it to set PTE_RDONLY. The failing sequence is:
1. We put down a PTE_WRITE | PTE_DIRTY | PTE_AF pte
2. Memory pressure -> pte_mkold(pte) -> clear PTE_AF
3. A read faults due to the missing access flag
4. ptep_set_access_flags is called with dirty = 0, due to the read fault
5. pte is then made PTE_WRITE | PTE_DIRTY | PTE_AF | PTE_RDONLY (!)
6. A write faults, but pte_write is true so we get stuck
The solution is to check the new page table entry (as would be done by
the generic, non-atomic definition of ptep_set_access_flags that just
calls set_pte_at) to establish the dirty state.
Fixes: 66dbd6e61a52 ("arm64: Implement ptep_set_access_flags() for hardware AF/DBM")
Reviewed-by: Catalin Marinas <catalin.marinas@arm.com>
Reported-by: Alexander Graf <agraf@suse.de>
Tested-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
commit 66dbd6e61a526ae7d11a208238ae2c17e5cacb6b upstream.
When hardware updates of the access and dirty states are enabled, the
default ptep_set_access_flags() implementation based on calling
set_pte_at() directly is potentially racy. This triggers the "racy dirty
state clearing" warning in set_pte_at() because an existing writable PTE
is overridden with a clean entry.
There are two main scenarios for this situation:
1. The CPU getting an access fault does not support hardware updates of
the access/dirty flags. However, a different agent in the system
(e.g. SMMU) can do this, therefore overriding a writable entry with a
clean one could potentially lose the automatically updated dirty
status
2. A more complex situation is possible when all CPUs support hardware
AF/DBM:
a) Initial state: shareable + writable vma and pte_none(pte)
b) Read fault taken by two threads of the same process on different
CPUs
c) CPU0 takes the mmap_sem and proceeds to handling the fault. It
eventually reaches do_set_pte() which sets a writable + clean pte.
CPU0 releases the mmap_sem
d) CPU1 acquires the mmap_sem and proceeds to handle_pte_fault(). The
pte entry it reads is present, writable and clean and it continues
to pte_mkyoung()
e) CPU1 calls ptep_set_access_flags()
If between (d) and (e) the hardware (another CPU) updates the dirty
state (clears PTE_RDONLY), CPU1 will override the PTR_RDONLY bit
marking the entry clean again.
This patch implements an arm64-specific ptep_set_access_flags() function
to perform an atomic update of the PTE flags.
Fixes: 2f4b829c625e ("arm64: Add support for hardware updates of the access and dirty pte bits")
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Reported-by: Ming Lei <tom.leiming@gmail.com>
Tested-by: Julien Grall <julien.grall@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
[will: reworded comment]
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
commit dfd55ad85e4a7fbaa82df12467515ac3c81e8a3e upstream.
Commit dd006da21646 ("arm64: mm: increase VA range of identity map") made
some changes to the memory mapping code to allow physical memory to reside
at an offset that exceeds the size of the virtual mapping.
However, since the size of the vmemmap area is proportional to the size of
the VA area, but it is populated relative to the physical space, we may
end up with the struct page array being mapped outside of the vmemmap
region. For instance, on my Seattle A0 box, I can see the following output
in the dmesg log.
vmemmap : 0xffffffbdc0000000 - 0xffffffbfc0000000 ( 8 GB maximum)
0xffffffbfc0000000 - 0xffffffbfd0000000 ( 256 MB actual)
We can fix this by deciding that the vmemmap region is not a projection of
the physical space, but of the virtual space above PAGE_OFFSET, i.e., the
linear region. This way, we are guaranteed that the vmemmap region is of
sufficient size, and we can even reduce the size by half.
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
commit 722ec35f7faefcc34d12616eca7976a848870f9d upstream.
This patch ensures that devices, which got registered before arch_initcall
will be handled correctly by IOMMU-based DMA-mapping code.
Fixes: 13b8629f6511 ("arm64: Add IOMMU dma_ops")
Acked-by: Robin Murphy <robin.murphy@arm.com>
Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
commit 57adec866c0440976c96a4b8f5b59fb411b1cacb upstream.
Calling apply_to_page_range with an empty range results in a BUG_ON
from the core code. This can be triggered by trying to load the st_drv
module with CONFIG_DEBUG_SET_MODULE_RONX enabled:
kernel BUG at mm/memory.c:1874!
Internal error: Oops - BUG: 0 [#1] PREEMPT SMP
Modules linked in:
CPU: 3 PID: 1764 Comm: insmod Not tainted 4.5.0-rc1+ #2
Hardware name: ARM Juno development board (r0) (DT)
task: ffffffc9763b8000 ti: ffffffc975af8000 task.ti: ffffffc975af8000
PC is at apply_to_page_range+0x2cc/0x2d0
LR is at change_memory_common+0x80/0x108
This patch fixes the issue by making change_memory_common (called by the
set_memory_* functions) a NOP when numpages == 0, therefore avoiding the
erroneous call to apply_to_page_range and bringing us into line with x86
and s390.
Reviewed-by: Laura Abbott <labbott@redhat.com>
Acked-by: David Rientjes <rientjes@google.com>
Signed-off-by: Mika Penttilä <mika.penttila@nextfour.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
commit f436b2ac90a095746beb6729b8ee8ed87c9eaede upstream.
The Performance Monitors extension is an optional feature of the
AArch64 architecture, therefore, in order to access Performance
Monitors registers safely, the kernel should detect the architected
PMU unit presence through the ID_AA64DFR0_EL1 register PMUVer field
before accessing them.
This patch implements a guard by reading the ID_AA64DFR0_EL1 register
PMUVer field to detect the architected PMU presence and prevent accessing
PMU system registers if the Performance Monitors extension is not
implemented in the core.
Cc: Peter Maydell <peter.maydell@linaro.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Fixes: 60792ad349f3 ("arm64: kernel: enforce pmuserenr_el0 initialization and restore")
Signed-off-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Reported-by: Guenter Roeck <linux@roeck-us.net>
Tested-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
commit 60792ad349f3c6dc5735aafefe5dc9121c79e320 upstream.
The pmuserenr_el0 register value is architecturally UNKNOWN on reset.
Current kernel code resets that register value iff the core pmu device is
correctly probed in the kernel. On platforms with missing DT pmu nodes (or
disabled perf events in the kernel), the pmu is not probed, therefore the
pmuserenr_el0 register is not reset in the kernel, which means that its
value retains the reset value that is architecturally UNKNOWN (system
may run with eg pmuserenr_el0 == 0x1, which means that PMU counters access
is available at EL0, which must be disallowed).
This patch adds code that resets pmuserenr_el0 on cold boot and restores
it on core resume from shutdown, so that the pmuserenr_el0 setup is
always enforced in the kernel.
Cc: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
commit 32d6397805d00573ce1fa55f408ce2bca15b0ad3 upstream.
In paging_init, we allocate the zero page, memset it to zero and then
point TTBR0 to it in order to avoid speculative fetches through the
identity mapping.
In order to guarantee that the freshly zeroed page is indeed visible to
the page table walker, we need to execute a dsb instruction prior to
writing the TTBR.
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
This reverts commit 348a65cdcbbf243073ee39d1f7d4413081ad7eab.
Incorrect page table manipulation that does not respect the ARM ARM
recommended break-before-make sequence may lead to TLB conflicts. The
contiguous PTE patch makes the system even more susceptible to such
errors by changing the mapping from a single page to a contiguous range
of pages. An additional TLB invalidation would reduce the risk window,
however, the correct fix is to switch to a temporary swapper_pg_dir.
Once the correct workaround is done, the reverted commit will be
re-applied.
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Reported-by: Jeremy Linton <jeremy.linton@arm.com>
|
|
Under some unusual context-switching patterns, it is possible to end up
with multiple threads from the same mm running concurrently with
different ASIDs:
1. CPU x schedules task t with mm p containing ASID a and generation g
This task doesn't block and the CPU doesn't context switch.
So:
* per_cpu(active_asid, x) = {g,a}
* p->context.id = {g,a}
2. Some other CPU generates an ASID rollover. The global generation is
now (g + 1). CPU x is still running t, with no context switch and
so per_cpu(reserved_asid, x) = {g,a}
3. CPU y schedules task t', which shares mm p with t. The generation
mismatches, so we take the slowpath and hit the reserved ASID from
CPU x. p is then updated so that p->context.id = {g + 1,a}
4. CPU y schedules some other task u, which has an mm != p.
5. Some other CPU generates *another* CPU rollover. The global
generation is now (g + 2). CPU x is still running t, with no context
switch and so per_cpu(reserved_asid, x) = {g,a}.
6. CPU y once again schedules task t', but now *fails* to hit the
reserved ASID from CPU x because of the generation mismatch. This
results in a new ASID being allocated, despite the fact that t is
still running on CPU x with the same mm.
Consequently, TLBIs (e.g. as a result of CoW) will not be synchronised
between the two threads.
This patch fixes the problem by updating all of the matching reserved
ASIDs when we hit on the slowpath (i.e. in step 3 above). This keeps
the reserved ASIDs in-sync with the mm and avoids the problem.
Reported-by: Tony Thompson <anthony.thompson@arm.com>
Reviewed-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
|
|
We are missing descriptions for some valid xFSC values in the fault info
table (e.g. "TLB conflict abort"), and have erroneous descriptions for
reserved values (e.g. "asynchronous external abort", "debug event").
This patch adds the missing xFSC values, and removes erroneous decoding
of values reserved by the architecture, as described in ARM DDI 0487A.h.
At the same time, fixed the unbalanced brackets for the synchronous
parity error strings in the table.
Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Reviewed-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
|
|
In early_alloc we check if the memblock_alloc failed by checking
the virtual address of the result, which will never fail. This patch
fixes it to check the actual result for failure.
Acked-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Suzuki K. Poulose <suzuki.poulose@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
|
|
The permissions in mark_rodata_ro trigger a build error
with STRICT_MM_TYPECHECKS. Fix this by introducing
PAGE_KERNEL_ROX for the same reasons as PAGE_KERNEL_RO.
From Ard:
"PAGE_KERNEL_EXEC has PTE_WRITE set as well, making the range
writeable under the ARMv8.1 DBM feature, that manages the
dirty bit in hardware (writing to a page with the PTE_RDONLY
and PTE_WRITE bits both set will clear the PTE_RDONLY bit in that case)"
Signed-off-by: Laura Abbott <labbott@fedoraproject.org>
Acked-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
|
|
Including linux/acpi.h from asm/dma-mapping.h causes tons of compile-time
warnings, e.g.
drivers/isdn/mISDN/dsp_ecdis.h:43:0: warning: "FALSE" redefined
drivers/isdn/mISDN/dsp_ecdis.h:44:0: warning: "TRUE" redefined
drivers/net/fddi/skfp/h/targetos.h:62:0: warning: "TRUE" redefined
drivers/net/fddi/skfp/h/targetos.h:63:0: warning: "FALSE" redefined
However, it looks like the dependency should not even there as
I do not see why __generic_dma_ops() cares about whether we have
an ACPI based system or not.
The current behavior is to fall back to the global dma_ops when
a device has not set its own dma_ops, but only for DT based systems.
This seems dangerous, as a random device might have different
requirements regarding IOMMU or coherency, so we should really
never have that fallback and just forbid DMA when we have not
initialized DMA for a device.
This removes the global dma_ops variable and the special-casing
for ACPI, and just returns the dma ops that got set for the
device, or the dummy_dma_ops if none were present.
The original code has apparently been copied from arm32 where we
rely on it for ISA devices things like the floppy controller, but
we should have no such devices on ARM64.
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
[catalin.marinas@arm.com: removed acpi_disabled check in arch_setup_dma_ops()]
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
|
|
When booting a 64k pages kernel that is built with CONFIG_DEBUG_RODATA
and resides at an offset that is not a multiple of 512 MB, the rounding
that occurs in __map_memblock() and fixup_executable() results in
incorrect regions being mapped.
The following snippet from /sys/kernel/debug/kernel_page_tables shows
how, when the kernel is loaded 2 MB above the base of DRAM at 0x40000000,
the first 2 MB of memory (which may be inaccessible from non-secure EL1
or just reserved by the firmware) is inadvertently mapped into the end of
the module region.
---[ Modules start ]---
0xfffffdffffe00000-0xfffffe0000000000 2M RW NX ... UXN MEM/NORMAL
---[ Modules end ]---
---[ Kernel Mapping ]---
0xfffffe0000000000-0xfffffe0000090000 576K RW NX ... UXN MEM/NORMAL
0xfffffe0000090000-0xfffffe0000200000 1472K ro x ... UXN MEM/NORMAL
0xfffffe0000200000-0xfffffe0000800000 6M ro x ... UXN MEM/NORMAL
0xfffffe0000800000-0xfffffe0000810000 64K ro x ... UXN MEM/NORMAL
0xfffffe0000810000-0xfffffe0000a00000 1984K RW NX ... UXN MEM/NORMAL
0xfffffe0000a00000-0xfffffe00ffe00000 4084M RW NX ... UXN MEM/NORMAL
The same issue is likely to occur on 16k pages kernels whose load
address is not a multiple of 32 MB (i.e., SECTION_SIZE). So round to
SWAPPER_BLOCK_SIZE instead of SECTION_SIZE.
Fixes: da141706aea5 ("arm64: add better page protections to arm64")
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Acked-by: Mark Rutland <mark.rutland@arm.com>
Acked-by: Laura Abbott <labbott@redhat.com>
Cc: <stable@vger.kernel.org> # 4.0+
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
|
|
The iommu-dma layer does its own size-alignment for coherent DMA
allocations based on IOMMU page sizes, but we still need to consider
CPU page sizes for the cases where a non-cacheable CPU mapping is
created. Whilst everything on the alloc/map path seems to implicitly
align things enough to make it work, some functions used by the
corresponding unmap/free path do not, which leads to problems freeing
odd-sized allocations. Either way it's something we really should be
handling explicitly, so do that to make both paths suitably robust.
Reported-by: Yong Wu <yong.wu@mediatek.com>
Signed-off-by: Robin Murphy <robin.murphy@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux
Pull arm64 fixes and clean-ups from Catalin Marinas:
"Here's a second pull request for this merging window with some
fixes/clean-ups:
- __cmpxchg_double*() return type fix to avoid truncation of a long
to int and subsequent logical "not" in cmpxchg_double()
misinterpreting the operation success/failure
- BPF fixes for mod and div by zero
- Fix compilation with STRICT_MM_TYPECHECKS enabled
- VDSO build fix without libgcov
- Some static and __maybe_unused annotations
- Kconfig clean-up (FRAME_POINTER)
- defconfig update for CRYPTO_CRC32_ARM64"
* tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux:
arm64: suspend: make hw_breakpoint_restore static
arm64: mmu: make split_pud and fixup_executable static
arm64: smp: make of_parse_and_init_cpus static
arm64: use linux/types.h in kvm.h
arm64: build vdso without libgcov
arm64: mark cpus_have_hwcap as __maybe_unused
arm64: remove redundant FRAME_POINTER kconfig option and force to select it
arm64: fix R/O permissions of FDT mapping
arm64: fix STRICT_MM_TYPECHECKS issue in PTE_CONT manipulation
arm64: bpf: fix mod-by-zero case
arm64: bpf: fix div-by-zero case
arm64: Enable CRYPTO_CRC32_ARM64 in defconfig
arm64: cmpxchg_dbl: fix return value type
|
|
split_pud and fixup_executable are only called from within mmu.c, so
they can be declared static.
Signed-off-by: Jisheng Zhang <jszhang@marvell.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
|
|
The mapping permissions of the FDT are set to 'PAGE_KERNEL | PTE_RDONLY'
in an attempt to map the FDT as read-only. However, not only does this
break at build time under STRICT_MM_TYPECHECKS (since the two terms are
of different types in that case), it also results in both the PTE_WRITE
and PTE_RDONLY attributes to be set, which means the region is still
writable under ARMv8.1 DBM (and an attempted write will simply clear the
PT_RDONLY bit).
So instead, define PAGE_KERNEL_RO (which already has an established
meaning across architectures) and use that instead.
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
|
|
The new page table code that manipulates the PTE_CONT flags does so
in a way that is inconsistent with STRICT_MM_TYPECHECKS. Fix it by
using the correct combination of __pgprot() and pgprot_val().
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
|
|
__GFP_WAIT was renamed for __GFP_RECLAIM and the gfpflags_allow_blocking()
helper was added.
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Robin Murphy <robin.murphy@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
|
|
sleep and avoiding waking kswapd
__GFP_WAIT has been used to identify atomic context in callers that hold
spinlocks or are in interrupts. They are expected to be high priority and
have access one of two watermarks lower than "min" which can be referred
to as the "atomic reserve". __GFP_HIGH users get access to the first
lower watermark and can be called the "high priority reserve".
Over time, callers had a requirement to not block when fallback options
were available. Some have abused __GFP_WAIT leading to a situation where
an optimisitic allocation with a fallback option can access atomic
reserves.
This patch uses __GFP_ATOMIC to identify callers that are truely atomic,
cannot sleep and have no alternative. High priority users continue to use
__GFP_HIGH. __GFP_DIRECT_RECLAIM identifies callers that can sleep and
are willing to enter direct reclaim. __GFP_KSWAPD_RECLAIM to identify
callers that want to wake kswapd for background reclaim. __GFP_WAIT is
redefined as a caller that is willing to enter direct reclaim and wake
kswapd for background reclaim.
This patch then converts a number of sites
o __GFP_ATOMIC is used by callers that are high priority and have memory
pools for those requests. GFP_ATOMIC uses this flag.
o Callers that have a limited mempool to guarantee forward progress clear
__GFP_DIRECT_RECLAIM but keep __GFP_KSWAPD_RECLAIM. bio allocations fall
into this category where kswapd will still be woken but atomic reserves
are not used as there is a one-entry mempool to guarantee progress.
o Callers that are checking if they are non-blocking should use the
helper gfpflags_allow_blocking() where possible. This is because
checking for __GFP_WAIT as was done historically now can trigger false
positives. Some exceptions like dm-crypt.c exist where the code intent
is clearer if __GFP_DIRECT_RECLAIM is used instead of the helper due to
flag manipulations.
o Callers that built their own GFP flags instead of starting with GFP_KERNEL
and friends now also need to specify __GFP_KSWAPD_RECLAIM.
The first key hazard to watch out for is callers that removed __GFP_WAIT
and was depending on access to atomic reserves for inconspicuous reasons.
In some cases it may be appropriate for them to use __GFP_HIGH.
The second key hazard is callers that assembled their own combination of
GFP flags instead of starting with something like GFP_KERNEL. They may
now wish to specify __GFP_KSWAPD_RECLAIM. It's almost certainly harmless
if it's missed in most cases as other activity will wake kswapd.
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Christoph Lameter <cl@linux.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Vitaly Wool <vitalywool@gmail.com>
Cc: Rik van Riel <riel@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/joro/iommu
Pull iommu updates from Joerg Roedel:
"This time including:
- A new IOMMU driver for s390 pci devices
- Common dma-ops support based on iommu-api for ARM64. The plan is
to use this as a basis for ARM32 and hopefully other architectures
as well in the future.
- MSI support for ARM-SMMUv3
- Cleanups and dead code removal in the AMD IOMMU driver
- Better RMRR handling for the Intel VT-d driver
- Various other cleanups and small fixes"
* tag 'iommu-updates-v4.4' of git://git.kernel.org/pub/scm/linux/kernel/git/joro/iommu: (41 commits)
iommu/vt-d: Fix return value check of parse_ioapics_under_ir()
iommu/vt-d: Propagate error-value from ir_parse_ioapic_hpet_scope()
iommu/vt-d: Adjust the return value of the parse_ioapics_under_ir
iommu: Move default domain allocation to iommu_group_get_for_dev()
iommu: Remove is_pci_dev() fall-back from iommu_group_get_for_dev
iommu/arm-smmu: Switch to device_group call-back
iommu/fsl: Convert to device_group call-back
iommu: Add device_group call-back to x86 iommu drivers
iommu: Add generic_device_group() function
iommu: Export and rename iommu_group_get_for_pci_dev()
iommu: Revive device_group iommu-ops call-back
iommu/amd: Remove find_last_devid_on_pci()
iommu/amd: Remove first/last_device handling
iommu/amd: Initialize amd_iommu_last_bdf for DEV_ALL
iommu/amd: Cleanup buffer allocation
iommu/amd: Remove cmd_buf_size and evt_buf_size from struct amd_iommu
iommu/amd: Align DTE flag definitions
iommu/amd: Remove old alias handling code
iommu/amd: Set alias DTE in do_attach/do_detach
iommu/amd: WARN when __[attach|detach]_device are called with irqs enabled
...
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux
Pull arm64 updates from Catalin Marinas:
- "genirq: Introduce generic irq migration for cpu hotunplugged" patch
merged from tip/irq/for-arm to allow the arm64-specific part to be
upstreamed via the arm64 tree
- CPU feature detection reworked to cope with heterogeneous systems
where CPUs may not have exactly the same features. The features
reported by the kernel via internal data structures or ELF_HWCAP are
delayed until all the CPUs are up (and before user space starts)
- Support for 16KB pages, with the additional bonus of a 36-bit VA
space, though the latter only depending on EXPERT
- Implement native {relaxed, acquire, release} atomics for arm64
- New ASID allocation algorithm which avoids IPI on roll-over, together
with TLB invalidation optimisations (using local vs global where
feasible)
- KASan support for arm64
- EFI_STUB clean-up and isolation for the kernel proper (required by
KASan)
- copy_{to,from,in}_user optimisations (sharing the memcpy template)
- perf: moving arm64 to the arm32/64 shared PMU framework
- L1_CACHE_BYTES increased to 128 to accommodate Cavium hardware
- Support for the contiguous PTE hint on kernel mapping (16 consecutive
entries may be able to use a single TLB entry)
- Generic CONFIG_HZ now used on arm64
- defconfig updates
* tag 'arm64-upstream' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux: (91 commits)
arm64/efi: fix libstub build under CONFIG_MODVERSIONS
ARM64: Enable multi-core scheduler support by default
arm64/efi: move arm64 specific stub C code to libstub
arm64: page-align sections for DEBUG_RODATA
arm64: Fix build with CONFIG_ZONE_DMA=n
arm64: Fix compat register mappings
arm64: Increase the max granular size
arm64: remove bogus TASK_SIZE_64 check
arm64: make Timer Interrupt Frequency selectable
arm64/mm: use PAGE_ALIGNED instead of IS_ALIGNED
arm64: cachetype: fix definitions of ICACHEF_* flags
arm64: cpufeature: declare enable_cpu_capabilities as static
genirq: Make the cpuhotplug migration code less noisy
arm64: Constify hwcap name string arrays
arm64/kvm: Make use of the system wide safe values
arm64/debug: Make use of the system wide safe value
arm64: Move FP/ASIMD hwcap handling to common code
arm64/HWCAP: Use system wide safe values
arm64/capabilities: Make use of system wide safe value
arm64: Delay cpu feature capability checks
...
|
|
Trying to build with CONFIG_ZONE_DMA=n leaves visible references
to the now-undefined ZONE_DMA, resulting in a syntax error.
Hide the references behind an #ifdef instead of using IS_ENABLED.
Signed-off-by: Robin Murphy <robin.murphy@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
|
|
The <linux/mm.h> already provides the PAGE_ALIGNED macro. Let's
use this macro instead of IS_ALIGNED and passing PAGE_SIZE directly.
Signed-off-by: Alexander Kuleshov <kuleshovmail@gmail.com>
Acked-by: Laura Abbott <laura@labbott.name>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
|
|
At the moment we run through the arm64_features capability list for
each CPU and set the capability if one of the CPU supports it. This
could be problematic in a heterogeneous system with differing capabilities.
Delay the CPU feature checks until all the enabled CPUs are up(i.e,
smp_cpus_done(), so that we can make better decisions based on the
overall system capability. Once we decide and advertise the capabilities
the alternatives can be applied. From this state, we cannot roll back
a feature to disabled based on the values from a new hotplugged CPU,
due to the runtime patching and other reasons. So, for all new CPUs,
we need to make sure that they have the established system capabilities.
Failing which, we bring the CPU down, preventing it from turning online.
Once the capabilities are decided, any new CPU booting up goes through
verification to ensure that it has all the enabled capabilities and also
invokes the respective enable() method on the CPU.
The CPU errata checks are not delayed and is still executed per-CPU
to detect the respective capabilities. If we ever come across a non-errata
capability that needs to be checked on each-CPU, we could introduce them via
a new capability table(or introduce a flag), which can be processed per CPU.
The next patch will make the feature checks use the system wide
safe value of a feature register.
NOTE: The enable() methods associated with the capability is scheduled
on all the CPUs (which is the only use case at the moment). If we need
a different type of 'enable()' which only needs to be run once on any CPU,
we should be able to handle that when needed.
Signed-off-by: Suzuki K. Poulose <suzuki.poulose@arm.com>
Tested-by: Dave Martin <Dave.Martin@arm.com>
[catalin.marinas@arm.com: static variable and coding style fixes]
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
|
|
This patch turns on the 16K page support in the kernel. We
support 48bit VA (4 level page tables) and 47bit VA (3 level
page tables).
With 16K we can map 128 entries using contiguous bit hint
at level 3 to map 2M using single TLB entry.
TODO: 16K supports 32 contiguous entries at level 2 to get us
1G(which is not yet supported by the infrastructure). That should
be a separate patch altogether.
Cc: Will Deacon <will.deacon@arm.com>
Cc: Jeremy Linton <jeremy.linton@arm.com>
Cc: Marc Zyngier <marc.zyngier@arm.com>
Cc: Christoffer Dall <christoffer.dall@linaro.org>
Cc: Steve Capper <steve.capper@linaro.org>
Signed-off-by: Suzuki K. Poulose <suzuki.poulose@arm.com>
Reviewed-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Tested-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Acked-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
|
|
We use section maps with 4K page size to create the swapper/idmaps.
So far we have used !64K or 4K checks to handle the case where we
use the section maps.
This patch adds a new symbol, ARM64_SWAPPER_USES_SECTION_MAPS, to
handle cases where we use section maps, instead of using the page size
symbols.
Cc: Will Deacon <will.deacon@arm.com>
Signed-off-by: Suzuki K. Poulose <suzuki.poulose@arm.com>
Reviewed-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Acked-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
|
|
With iommu_dma_ops in place, hook them up to the configuration code, so
IOMMU-fronted devices will get them automatically.
Acked-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Robin Murphy <robin.murphy@arm.com>
Signed-off-by: Joerg Roedel <jroedel@suse.de>
|
|
Taking some inspiration from the arch/arm code, implement the
arch-specific side of the DMA mapping ops using the new IOMMU-DMA layer.
Since there is still work to do elsewhere to make DMA configuration happen
in a more appropriate order and properly support platform devices in the
IOMMU core, the device setup code unfortunately starts out carrying some
workarounds to ensure it works correctly in the current state of things.
Signed-off-by: Robin Murphy <robin.murphy@arm.com>
Acked-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Joerg Roedel <jroedel@suse.de>
|
|
Signed-off-by: Ingo Molnar <mingo@kernel.org>
|
|
Sparse reports some new issues introduced by the kasan patches:
arch/arm64/mm/kasan_init.c:91:13: warning: no previous prototype for
'kasan_early_init' [-Wmissing-prototypes] void __init kasan_early_init(void)
^
arch/arm64/mm/kasan_init.c:91:13: warning: symbol 'kasan_early_init'
was not declared. Should it be static? [sparse]
This patch resolves the problem by adding a prototype for
kasan_early_init and marking the function as asmlinkage, since it's only
called from head.S.
Signed-off-by: Will Deacon <will.deacon@arm.com>
Acked-by: Andrey Ryabinin <ryabinin.a.a@gmail.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
|
|
This prints out the virtual memory assigned to KASan in the
boot crawl along with other memory assignments, if and only
if KASan is activated.
Example dmesg from the Juno Development board:
Memory: 1691156K/2080768K available (5465K kernel code, 444K rwdata,
2160K rodata, 340K init, 217K bss, 373228K reserved, 16384K cma-reserved)
Virtual kernel memory layout:
kasan : 0xffffff8000000000 - 0xffffff9000000000 ( 64 GB)
vmalloc : 0xffffff9000000000 - 0xffffffbdbfff0000 ( 182 GB)
vmemmap : 0xffffffbdc0000000 - 0xffffffbfc0000000 ( 8 GB maximum)
0xffffffbdc2000000 - 0xffffffbdc3fc0000 ( 31 MB actual)
fixed : 0xffffffbffabfd000 - 0xffffffbffac00000 ( 12 KB)
PCI I/O : 0xffffffbffae00000 - 0xffffffbffbe00000 ( 16 MB)
modules : 0xffffffbffc000000 - 0xffffffc000000000 ( 64 MB)
memory : 0xffffffc000000000 - 0xffffffc07f000000 ( 2032 MB)
.init : 0xffffffc0007f5000 - 0xffffffc00084a000 ( 340 KB)
.text : 0xffffffc000080000 - 0xffffffc0007f45b4 ( 7634 KB)
.data : 0xffffffc000850000 - 0xffffffc0008bf200 ( 445 KB)
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Andrey Ryabinin <ryabinin.a.a@gmail.com>
Acked-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
|
|
This patch adds arch specific code for kernel address sanitizer
(see Documentation/kasan.txt).
1/8 of kernel addresses reserved for shadow memory. There was no
big enough hole for this, so virtual addresses for shadow were
stolen from vmalloc area.
At early boot stage the whole shadow region populated with just
one physical page (kasan_zero_page). Later, this page reused
as readonly zero shadow for some memory that KASan currently
don't track (vmalloc).
After mapping the physical memory, pages for shadow memory are
allocated and mapped.
Functions like memset/memmove/memcpy do a lot of memory accesses.
If bad pointer passed to one of these function it is important
to catch this. Compiler's instrumentation cannot do this since
these functions are written in assembly.
KASan replaces memory functions with manually instrumented variants.
Original functions declared as weak symbols so strong definitions
in mm/kasan/kasan.c could replace them. Original functions have aliases
with '__' prefix in name, so we could call non-instrumented variant
if needed.
Some files built without kasan instrumentation (e.g. mm/slub.c).
Original mem* function replaced (via #define) with prefixed variants
to disable memory access checks for such files.
Signed-off-by: Andrey Ryabinin <ryabinin.a.a@gmail.com>
Tested-by: Linus Walleij <linus.walleij@linaro.org>
Reviewed-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
|
|
This will be used by KASAN latter.
Signed-off-by: Andrey Ryabinin <ryabinin.a.a@gmail.com>
Acked-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
|
|
For more control over which functions are called with the MMU off or
with the UEFI 1:1 mapping active, annotate some assembler routines as
position independent. This is done by introducing ENDPIPROC(), which
replaces the ENDPROC() declaration of those routines.
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
|
|
With 64k pages, the next larger segment size is 512M. The linux
kernel also uses different protection flags to cover its code and data.
Because of this requirement, the vast majority of the kernel code and
data structures end up being mapped with 64k pages instead of the larger
pages common with a 4k page kernel.
Recent ARM processors support a contiguous bit in the
page tables which allows the a TLB to cover a range larger than a
single PTE if that range is mapped into physically contiguous
ram.
So, for the kernel its a good idea to set this flag. Some basic
micro benchmarks show it can significantly reduce the number of
L1 dTLB refills.
Add boot option to enable/disable CONT marking, as well as fix a
bug found by Steve Capper.
Signed-off-by: Jeremy Linton <jeremy.linton@arm.com>
[catalin.marinas@arm.com: remove CONFIG_ARM64_CONT_PTE altogether]
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
|
|
The kernel page dump utility needs to be aware of the CONT bit before
it will break up pages ranges for display.
Signed-off-by: Jeremy Linton <jeremy.linton@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
|
|
mm_cpumask isn't actually used for anything on arm64, so remove all the
code trying to keep it up-to-date.
Reviewed-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
|
|
switch_mm performs some checks to try and avoid entering the ASID
allocator:
(1) If we're switching to the init_mm (no user mappings), then simply
set a reserved TTBR0 value with no page table (the zero page)
(2) If prev == next *and* the mm_cpumask indicates that we've run on
this CPU before, then we can skip the allocator.
However, there is plenty of redundancy here. With the new ASID allocator,
if prev == next, then we know that our ASID is valid and do not need to
worry about re-allocation. Consequently, we can drop the mm_cpumask check
in (2) and move the prev == next check before the init_mm check, since
if prev == next == init_mm then there's nothing to do.
Reviewed-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
|
|
Our current switch_mm implementation suffers from a number of problems:
(1) The ASID allocator relies on IPIs to synchronise the CPUs on a
rollover event
(2) Because of (1), we cannot allocate ASIDs with interrupts disabled
and therefore make use of a TIF_SWITCH_MM flag to postpone the
actual switch to finish_arch_post_lock_switch
(3) We run context switch with a reserved (invalid) TTBR0 value, even
though the ASID and pgd are updated atomically
(4) We take a global spinlock (cpu_asid_lock) during context-switch
(5) We use h/w broadcast TLB operations when they are not required
(e.g. in flush_context)
This patch addresses these problems by rewriting the ASID algorithm to
match the bitmap-based arch/arm/ implementation more closely. This in
turn allows us to remove much of the complications surrounding switch_mm,
including the ugly thread flag.
Reviewed-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
|
|
There are a number of places where a single CPU is running with a
private page-table and we need to perform maintenance on the TLB and
I-cache in order to ensure correctness, but do not require the operation
to be broadcast to other CPUs.
This patch adds local variants of tlb_flush_all and __flush_icache_all
to support these use-cases and updates the callers respectively.
__local_flush_icache_all also implies an isb, since it is intended to be
used synchronously.
Reviewed-by: Catalin Marinas <catalin.marinas@arm.com>
Acked-by: David Daney <david.daney@cavium.com>
Acked-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
|
|
When cold-booting a CPU, we must invalidate any junk entries from the
local TLB prior to enabling the MMU. This doesn't require broadcasting
within the inner-shareable domain, so de-scope the operation to apply
only to the local CPU.
Acked-by: Catalin Marinas <catalin.marinas@arm.com>
Reviewed-by: Catalin Marinas <catalin.marinas@arm.com>
Tested-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
|
|
This is the arm64 portion of commit 45cac65b0fcd ("readahead: fault
retry breaks mmap file read random detection"), which was absent from
the initial port and has since gone unnoticed. The original commit says:
> .fault now can retry. The retry can break state machine of .fault. In
> filemap_fault, if page is miss, ra->mmap_miss is increased. In the second
> try, since the page is in page cache now, ra->mmap_miss is decreased. And
> these are done in one fault, so we can't detect random mmap file access.
>
> Add a new flag to indicate .fault is tried once. In the second try, skip
> ra->mmap_miss decreasing. The filemap_fault state machine is ok with it.
With this change, Mark reports that:
> Random read improves by 250%, sequential read improves by 40%, and
> random write by 400% to an eMMC device with dm crypto wrapped around it.
Cc: Shaohua Li <shli@kernel.org>
Cc: Rik van Riel <riel@redhat.com>
Cc: Wu Fengguang <fengguang.wu@intel.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Mark Salyzyn <salyzyn@android.com>
Signed-off-by: Riley Andrews <riandrews@android.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
|
|
If CMA is turned on and CMA size is set to zero, kernel should
behave as if CMA was not enabled at compile time.
Every dma allocation should check existence of cma area
before requesting memory.
Arm has done this by commit e464ef16c4f0 ("arm: dma-mapping: add
checking cma area initialized"), also do this for arm64.
Acked-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Jisheng Zhang <jszhang@marvell.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
|
|
We don't want to expose the DCC to userspace, particularly as there is
a kernel console driver for it.
This patch resets mdscr_el1 to disable userspace access to the DCC
registers on the cold boot path.
Signed-off-by: Will Deacon <will.deacon@arm.com>
|
